A case-control study of injuries arising from the earthquake in Armenia, 1988
H.K. Armenian, E.K. Noji, & A.P. Oganesian.
Bulletin of the World Health Organization, 70(2): 251-257 (1992)
The study attempts to identify predictors of injuries among persons who were hospitalized following the Armenian earthquake of 7 December 1988. A total of 189 such individuals were identified through neighbourhood polyclinics in the city of Leninakan and 159 noninjured controls were selected from the same neighbourhoods. A standardized interview questionnaire was used. Cases and controls shared many social and demographic characteristics; however, 98% of persons who were hospitalized with injuries were inside a building at the time of the earthquake, compared with 83% of the controls (odds ratio = 12.20, 95% confidence interval (Cl) = 3.62-63.79). The odds ratio of injuries for individuals who were in a building that had five or more floors, compared with those in lower buildings, was 3.65 (95% Cl = 2.12-6.33). Leaving buildings after the first shock of the earthquake was a protective behaviour. The odds ratio for those staying indoors compared with those who ran out was 4.40 (95% Cl = 2.24-8.71).
2. H.K. Armenian et al.
of building, location of occupants), occupant beha- health care following the earthquake, relief efforts
viour, and personal characteristics in earthquake- and general health-risk behaviour. Each of the ques-
related injuries. tionnaires was coded and put in computer format for
processing and analysis. After simple frequency dis-
tribution analyses, the data for the cases were com-
Methods pared with those for the controls and odds ratios and
confidence intervals were calculated. The analysis
The study was begun in March 1989 and the data was unmatched, since a number of the cases were
collection phase was finished in September 1989. not matched to the controls, many of whom turned
The study population was defined as all persons who out to have been injured but not hospitalized.
were hospitalized with injuries from the city of Multivariate analyses were performed using logistic
Leninakan as a result of the earthquake of 7 regression.
December 1988. A list of these patients, collected
from discharge data from all the hospitals in
Armenia, was available from the Ministry of Health, Results
in Yerevan. However, case identification on the basis
of the addresses given on the list proved to be impos- A number of the social and demographic characteris-
sible since the residences of most of the patients with tics of the cases and controls were similar. Thus,
injuries had been destroyed and they were living in there were no significant differences between the
makeshift housing or moved frequently to different two groups in terms of age, sex, distribution and
places of residence. Because of these difficulties, the educational background. Similarly, there were no
selection of cases and controls was limited to identi- differences for factors related to general health be-
fication of individuals based on information provided haviour, such as smoking, exercise, and alcohol
by the neighbourhood polyclinics (primary care faci- consumption (Table 1).
lities which provide the entry point to the health ser- As shown in Table 2, 98.4% of the cases were
vices). From detailed maps of the city of Leninakan, inside a building at the time of the earthquake, com-
a random city block was selected on each of the days pared with 83.0% of the controls, with an odds ratio
of interviewing, and the neighbourhood polyclinic of 12.20 and a 95% confidence interval of
for the block concemed was contacted to identify the 3.62-63.79. The odds ratio for injuries for persons
cases of hospitalized injuries caused by the earthqu- who were in a building with five or more floors com-
ake. The name of each of the cases identified from pared to persons in lower buildings was 3.65 (range,
these neighbourhood searches was checked against 2.12-6.33) (Table 3). Also, the risk increased with
hospital discharge lists. For each of the cases, a
control who had not been hospitalized with injuries
was selected from another household in the same Table 1: Distribution of the study groups, according to
various characteristics
neighbourhood. An effort was made to match the
controls and cases by sex and age within ±5 years. No. of No. of controls with:
Since most of the interviewers, who were students, hospitalized
were employed for the summer months only, collec- cases No injuries Mild injuries
tion of data stopped in the autumn when the univer- Females 120 (63.5) a 95 (59.8) 42 (62.7)
sities re-opened. Interviews of 189 cases and 227 Educational level
controls were completed. However, 68 of the persons <High school 29 (19.7) 21 (15.2) 14 (23.3)
interviewed as controls had histories of injuries cau- High school 50 (34.0) 56 (40.6) 28 (46.7)
sed by the earthquake, but which did not necessitate Technical college 40 (27.2) 31 (22.5) 7 (11.7)
hospitalization. These were studied as a separate University 27 (19.6) 21 (14.3) 10 (16.7)
group. Less than 1% of the total study population Age group (years)
refused to be interviewed. After the variables of <17 35 (18.5) 21 (13.2) 5 ( 7.4)
interest had been defined, a set of questions was for- 17-22 23 (12.2) 12 ( 7.6) 10 (14.7)
mulated and a questionnaire was developed in 23-39 67 (35.5) 62 (39.0) 16 (23.5)
Armenian and pre-tested on a sample of 14 cases and 40-59 54 (28.6) 48 (30.2) 27 (39.7)
controls. The final questionnaire consisted of 67 >60 10 ( 5.3) 16 (10.1) 10 (14.7)
questions and took about 30 minutes to administer. Smokers 42 (22.2) 36 (22.6) 14 (20.6)
The questions covered sociodemographic informa- Alcohol consumption 40 (21.1) 46 (29.1) 10 (14.7)
tion about the person and the family, where they Regular exercise 81 (43.3) 59 (37.3) 22 (32.9)
were located and the position they adopted during Total 189 159 68
the earthquake, injuries, rescue activities, details of a
Figures in parentheses are percentages.
252 WHO Bulletin OMS. Vol 70 1992
3. Earthquake injuries In Armenia
Table 2: Distribution of cases and controls, according controls who had moved at all after the first shock.
to their location at the moment of the earthquake For this subgroup, the odds ratio for those who
No. of hospitalized No. of non-injured moved within the building compared with those who
Location cases controls ran out of the building was 3.84 (range, 1.77-8.42).
Table 6 shows the odds ratios after a multivaria-
In a building 179 (98.4) 132 (83.0) te adjustment of these findings using logistic regres-
In the street 3 ( 1.7) 27 (17.9) sion. The results indicate that this had no appreciable
Total 182 (100) 159 (100) effect, except for location on the upper floors of the
a Figures in parentheses are percentages. buildings, where the adjustment decreased the
magnitude of the association.
Table 5: Distribution of cases and controls, according
Table 3: Distribution of cases and controls, according to their reaction after the first shock of the earthquake
to the number of floors In the building they occupied at No. of No.of
the moment of the earthquake hospitalized non-injured
No. of hospitalized No. of non-injured cases controls
No. of floors cases controls Stayed indoorsa 144 (90.0) b 88 (67.2)
Ran or jumped outdoors 16 (10.0) 43 (32.8)
<4 91 (50.8) a 102 (79.1)
.5 88 (49.2) 27 (20.9) Total 160 (100 ) 131 (100)
Total 179 (100) 129 (100) a
Odds ratio for staying indoors = 4.40 (95% confidence inter-
val: 2.24-8.71).
a
Figures in parentheses are percentages. b Figures in parentheses are percentages.
Table 4: Distribution of cases and controls by location,
according to the floor they occupied in the building at Table 6: Multiple logistic regression analysis, adjusting
the moment of the earthquake for age group and the variables shown
No. of hospitalized No. of non-injured 95% Confidence
No. of floorsa cases controls Variable Odds ratio interval
.5 floors in the building 3.45 1.76-6.74
1 30 (16.8) b 60 (46.9) Individuals located on floors:
2-4 121 (67.6) 63 (49.2) 2-4 versus 1 2.60 1.42-4.75
.5 28 (15.6) 5 ( 3.9) 5 versus 1 4.02 2.08-14.9
Total 179 (100) 128 (100) Stayed indoors versus ran outside 4.84 2.34-10.0
a Odds ratios: floors 2-4 versus floor 1 = 3.84 (95% confidence
interval (Cl): 2.18-6.79); floors .5 versus floor 1 = 11.20 (95%
Cl: 3.62-37.03); Mantel-Haenszel weighted odds ratio = 4.95
(95% Cl: 3.04-8.16).
b Figures in parentheses are percentages. Discussion
Previous studies have stressed the importance of cri-
tically analysing earthquake data in order to develop
the number of the floor in the building (Table 4). methods for rapidly assessing health care needs and
The odds ratio for individuals who were on the improving disaster relief (1-5, 24, 25). In the after-
second to fourth floors at the time of the earthquake, math of the Armenian disaster, an opportunity arose
rather than on the first floor, was 3.84, while the for us to study the injuries in the city of Leninakan
odds ratio for those on the fifth floor or higher, and to relate them to different types of buildings,
rather than on the first floor, was 11.20. occupant locations, and occupant behaviours.
The first reaction of individuals who were There have been some attempts to correlate
within a building at the moment of the earthquake, death and injuries in earthquakes with structural fac-
after the initial shock, was to run outdoors as a pro- tors (6-9),d housing damage, victim behaviour
tective measure (odds ratio for staying indoors, 4.40;
(range, 2.24-8.71)) (Table 5). A separate analysis
was performed for the subgroup of cases and d See footnote a, p. 251.
WHO Bulletin OMS. Vol 70 1992 253
4. H.K. Armenian et al.
(10-14), and other possible determinants of injury. rather than protect building occupants (10).
Most of the previously studied earthquakes, how- Confusion about recommended first actions might
ever, occurred in rural areas, and except for the ear- arise because the relative efficacy of protective occu-
thquake in Mexico City in 1985, little research has pant actions is very much dependent on the seismic
been conducted on a major catastrophic earthquake performance of specific building types (13).
within an urban environment (15-19, 26, 27). Also, Widely-accepted beliefs about how people react
most studies have examined mortality and not mor- in earthquakes need to be re-examined; depending on
bidity. The present is the first case-control study the physical setting, some beliefs appear valid, others
that has focused on the determinants of injury in a potentially fatal. Experience indicates that different
major urban earthquake. types of building withstand earthquake shaking in
Among the factors that determine the number of quite dissimilar ways, thereby threatening occupants
people killed after a building collapses are entrap- in different ways.e Thus, in San Francisco and
ment, the severity of their injuries, how long indivi- Tokyo, whose buildings have a low probability of
duals can survive without medical attention, the time catastrophic collapse due to strict enforcement of
taken to rescue them and the medical treatment they building codes, and where there is some crowding of
receive (28, 29).e A 1977 study of the Guatemalan structures especially in central areas, the recommen-
earthquake concluded that deaths and injuries are cri- dation may be to stay indoors to avoid being struck
tically dependent on the damage to housing and on by a rain of bricks on to pavements from unbraced
the type of construction materials used (7). The most balconies or omamental parapets (12, 32-36). On the
important findings of this study related to the signifi- other hand, in countries with poor or nonexistent
cant differences between hospitalized cases and building codes, where the majority of the structures
uninjured controls in terms of their location at the have the potential for catastrophic collapse, e.g., in
time of the earthquake and their behaviour immedia- Armenia, running out of doors at the first instant of
tely after the impact. the earthquake may represent the only chance of sur-
The most common cause of earthquake injury in vival; however, this may be too simplistic. For
Armenia resulted from vibration-induced failure of example, in coastal towns of Peru that were affected
buildings, which entrapped victims (30). Such struc- by the catastrophic 1970 earthquake, people who
tures ranged from simple dwellings, with heavy roof- rushed instinctively out into the wide streets at the
ing material, to large multistorey buildings which first tremors escaped unscathed, and many of those
collapsed totally (22, 31). The structural engineering trapped in the collapsed houses with flimsy roofs
factors that lead to injury need to be addressed by were rescued (37). In contrast, the people in the
improving building design and construction practices mountainous Callejon de Huaylas who reacted simi-
and by stringently implementing the building codes; larly to the first tremors were immediately buried in
also, initial occupant reaction during an earthquake the narrow streets by tons of rubble bursting out
could be taught and modified (10). Often it is as- from both sides under the weight of heavy town
sumed that during a major earthquake the ground roofs. Also, in the parish of Venzone, hit by the 1976
motion will be too violent or the time will be too Friuli earthquake in northem Italy, agile groups suf-
short for building occupants to pursue any actions fered more than the elderly or very young because
that could affect their own survival or that of others. they ran out into the narrow streets and were crushed
However, the few studies that have appeared suggest by falling masonry (38). Clearly the development of
that occupants often have both the time and ability to sound guidelines indicating the best actions to take
take several actions before buildings collapse and to reduce the likelihood of injury will require further
that proper occupant action can be a decisive factor careful studies of the location of injured and non-
in either causing or preventing injury (10, 12). injured persons, correlated with specific geographi-
Education and training could therefore play a signifi- cal sites and building stock.
cant role in reducing earthquake morbidity and mor- In Armenia, retrospective comparison of the
tality; initial occupant actions following the first behaviour of hospitalized, injured, and noninjured
shock should, however, be based on a realistic controls suggests that running out of the building
appraisal of occupants' capabilities and actions decreased the injury rate. However, it is possible that
during earthquakes. Studies suggest that many gene- many of the cases were unable to run out of the buil-
ral beliefs about appropriate response can endanger ding because of their injuries, i.e., the injury itself
could have influenced their behaviour. Our separate
e Coburn, A.W. et al. Factors affecting fatalities and injury
analysis of the subgroup of cases and controls who
in earthquakes. Internal Report, Engineering Seismology and had moved at all after the first shock showed that
Earthquake Disaster Prevention Planning. Hokkaido, Japan, those who were able to move and stayed in the
Hokkaido University, 1987, pp. 1-80. building had an odds ratio of 3.84 for hospitalized
254 WHO Bulletin OMS. Vol 70 1992
5. Earthquake injuries in Armenia
injuries compared with those who ran out of the Study limitations
building. In Artnenia, the early waming offered
by preliminary tremors may also have been helpful in Although, the selection of the cases and controls in
decreasing earthquake morbidity. Similarly, an earth- the study was far from ideal, it reflects some of the
quake in the Yugoslavian state of Montenegro in difficulties encountered in random sampling
1979 came in two shocks, with enough time between approaches in communities that have been almost
them for people to get outside their houses (39). totally destroyed by a disaster. However, comparison
Although a building may still fail in an earth- of the demographic characteristics of the cases with
quake, injuries can be prevented or reduced if those those of the total population of hospitalized patients
parts of the building likely to be occupied by large from Leninakan indicated that for both groups a
numbers of people are designed in such a way that number of social and demographic factors were simi-
there is less risk of injury to the occupants. For lar, which argues against selection or sampling bias
example, one recommendation may be for structural for cases.
engineers to design safe emergency exits, especially The possibility of interviewer bias also has to be
in taller buildings. However, in order to design an considered since the case-control status of the study
effective escape route, both behavioural and engine- subjects was known to the interviewers. However,
ering factors need to be taken into consideration. those controls who tumed out to have minor injuries
Unfortunately, in many buildings the staircase is the upon interview had characteristics that were more
weakest element when the ground shakes (39). Thus, similar to those of the case group than to those of the
high-rise apartment complexes are very dangerous in controls. This speaks against interviewer bias.
earthquakes, since they have few emergency exits, Finally, there were no differences between the case
and vulnerable lifts and stairs. Clearly, proper occu- and control groups for a large number of variables
pant behaviour in earthquakes needs to be studied that may have been thought to be earthquake-related
further in multidisciplinary investigations involving but which were not.
epidemiologists and structural engineers (40, 41). Cultural factors and differences in building
Building types and structural systems have dif- materials, practices, and pattems vary significantly in
ferent collapse mechanisms when they fail under the different countries, as do pattems of building use.
influence of earthquake ground motion, and the type Care must therefore be exercised in generalizing the
of building and collapse pattem affect occupant sur- results of the present study to other countries, parti-
vival rates (28, 42). For example, in buildings that cularly to industrialized countries (23). Despite these
collapse quickly or that take a long time to evacuate limitations, the results of the study have increased
a high proportion of occupants may be trapped. The our understanding of the magnitude of the threats
few statistics that are available on reinforced con- that different building types and occupant behaviours
crete buildings show that about 90% of occupants pose and have suggested practical guidelines to
are trapped by collapse of these structures! This is improve public awareness and protection in earth-
probably due primarily to the difficulty of escaping quake-prone parts of the world, e.g., educational pro-
from multistorey buildings. These observations grammes that focus on protective actions to take in
are consistent with the results of our study, which earthquakes.
showed that persons located on the lower floors of
multistorey buildings were able to escape injury, or
at least severe injury that required hospitalization. Conclusions
Thus, not only are the type of building and the Our findings contribute to the overall process of
occupant's behaviour important determinants of mor- developing a methodology for the investigation of
bidity, but so also is the location of the victim at the the health effects of earthquakes. Studies such as this
time of the impact. Determining where people were should help to guide building construction practices
located when they were injured or killed can provide in earthquake-prone regions, suggest actions to be
valuable information to search and rescue per- taken by occupants to prevent death and injury, and
sonnel looking for potential survivors. provide insights that will lead to the development of
better earthquake-preparedness plans.
f
Acknowledgements
Spence, R.J.S. et al. Reducing human casualties in building
collapse: methods of optimising disaster plans to reduce injury This study was supported by a grant from the Armenian
levels. Cambridge University. Martin Centre for Architectural and Relief Society, Inc. Watertown, MA, USA. We acknowl-
Urban Studies, 1991, p. 25. edge the contributions of Arpi Balian, Nancy Rogers,
WHO Bulletin OMS. Vol 70 1992 255
6. H.K. Armenian et al.
Aaron Schechter, Karen Tuder, Mary Rybczynski and the of the International Workshop on Earthquake Injury
staff of the Republican Information and Computer Centre, Epidemiology: Implications for Mitigation and
Ministry of Health, Armenia, USSR. Response, Baltimore, MD, 10-12 July 1989.
Baltimore, MD, Johns Hopkins University, 1989, pp.
415-419.
Resume 2. Logue, J.N. et al. Research issues and directions
in the epidemiology of health effects of disasters.
Epidemiologic reviews, 3: 140-162 (1981).
Etude cas-temoins des blessures 3. Saidi, F. The 1962 earthquake in Iran. Some medi-
cons6cutives au tremblement de terre cal and social aspects. New England journal of
medicine, 268: 929-932 (1963).
d'Armnnie, 1988 4. Kates, R.W. et al. Human impact of the Managua
Dans cette etude, les auteurs ont cherch6 a earthquake. Science, 182: 981-990 (1973).
5. Whittaker, R. et al. Earthquake disaster in
d6terminer les facteurs predictifs de blessures Nicaragua: reflections on the initial management of
chez les personnes hospitalis6es a la suite du massive casualties. Journal of trauma, 14: 37-43
tremblement de terre survenu en' Arm6nie le 7 (1974).
d6cembre 1988. Ont ete retenus comme cas 189 6. de Bruycker, M. et al. The 1980 earthquake in
bless6s hospitalis6s dans un des dispensaires de southern Italy: rescue of trapped victims and morta-
quartier de L6ninakan, et comme temoins non lity. Bulletin of the World Health Organization, 61:
blesses 150 personnes provenant des memes 1021-1025 (1983).
7. Glass, R.l. et al. Earthquake injuries related to
quartiers. Un questionnaire par interview normali- housing in a Guatemalan village. Science, 197:
se a 6t6 utilise. Les cas et les t6moins pr6sen- 638-643 (1977).
taient de nombreuses caracteristiques sociales et 8. Mochiquki, T. et al. Human behavior and casualties
d6mographiques communes; toutefois, 98% des in wooden houses with little ductility. In: Proceed-
blesses hospitalises se trouvaient a l'interieur ings of Ninth World Conference on Earthquake
d'un batiment au moment du tremblement de Engineering, 8: 983-88 (1988).
terre, contre 83% des t6moins (odds ratio = 9. Lechat, M.F. Corporal damage as related to buil-
ding structure and design: the need for an interna-
12,20, intervalle de confiance a 95% (CI) = tional survey. In: Jones, N.P. et al., ed. Proceedings
3,62-63,79). Le odds ratio des blessures pour les of the International Workshop on Earthquake Injury
sujets se trouvant dans un batiment ayant cinq Epidemiology: Implications for Mitigation and
6tages ou plus, par rapport a ceux qui se trou- Response. Baltimore, MD, Johns Hopkins University
vaient dans des batiments moins hauts, 6tait de Press, 1989, pp. 1-16.
3,65 (Cl 95% = 2,12-6,33). Le risque de bles- 10. Durkin, M.E. Behavior of building occupants in
sures etait significativement plus eleve pour les earthquakes. Earthquake spectra, 1: 271-283
personnes qui se trouvaient dans les etages les (1985).
11. Lomnitz, C. Casualties and the behavior of popula-
plus elev6s lors du tremblement de terre. Le fait tions during earthquakes. Bulletin of the
de quitter les batiments des la premiere secousse Seismological Society of America, 60: 1309-1313
constituait un comportement protecteur. Le odds (1970).
ratio pour les sujets restes a l'int6rieur par rap- 12. Arnold, C. et al. Imperial County Services Building:
port a ceux qui sont sortis immediatement 6tait occupant behavior and operational consequences
as a result of the 1979 Imperial Valley earthquake.
de 4,40 (Cl 95% = 2,24-8,71). San Mateo, CA, August 1982, pp. 1-21 (National
Les resultats de cette 6tude 6pid6miologique Science Foundation Monograph).
contribuent au processus d'elaboration d'une 13. Aroni, S. & Durkin, M. Injuries and occupant beha-
methodologie d'investigation des effets des trem- vior in earthquakes. In: Joint US-Romanian Seminar
blements de terre sur la sante. De telles etudes on Earthquakes and Energy. Washington, DC,
aideront a orienter les pratiques de construction Architectural Research Centers Consortium, 1985,
dans les regions sismiques, a proposer des com- pp. 3-40.
14. de Bruycker, M. et al. The 1980 earthquake in
portements permettant aux occupants d'eviter southern Italy: morbidity and mortality. International
d'etre bless6s ou tues, et fourniront des indica- journal of epidemiology, 14: 113-117 (1985).
tions qui permettront d'am6liorer les plans d'inter- 15. de Ville de Goyet, C. & Jeannee, E. Epidemiological
vention en cas de s6isme en tenant compte des data on morbidity and mortality following the
types de constructions usit6s dans la r6gion. Guatemala earthquake. IRCS medical science: social
and occupational medicine, 4: 212 (1976).
16. Zeballos, J.L. Health effects of the Mexico earth-
quake - 19 September 1985. Disasters, 10: 141-
References 149 (1986).
1. Steinbrugge, K.V. Reducing earthquake casualties: 17. Chatterson, J. Guatemala after the earthquake.
an overview. In: Jones, N.P. et al., ed. Proceedings Canadian journal of public health, 67: 192-195
256 WHO Bulletin OMS. Vol 70 1992
7. Earthquake injuries in Armenia
(1976). cine, 19: 891-897 (1990).
18. Spencer, H.C., et al. Disease surveillance and deci- 31. Wyllie, L.A. & Lew, H.S. Performance of engine-
sion-making after the 1976 Guatemala earthquake. ered structures. Earthquake spectra (supplement),
Lancet, 2: 181 (1977). 70-92 (1989).
19. Sanchez-Carrillo, C.l. Morbidity following Mexico 32. Jones, N.P. et al. Preliminary earthquake injury epi-
City's 1985 earthquakes: clinical and epidemiologic demiology report: In: Bolin, R., ed. The Loma Prieta
findings from hospitals and emergency units. Public earthquake: studies of short-term impacts. Boulder,
health reports, 104: 483-488 (1989). CO, University of Colorado, 1990, pp. 33-43
20. Binder, S. & Sanderson, L.M. The role of the epi- (Natural Hazards Center Monograph).
demiologist in natural disasters. Annals of emergen- 33. Durkin, M.E. et al. Injuries in the Coalinga earth-
cy medicine, 16: 1081-1084 (1987). quake. In: The Coalinga earthquake of May 2, 1983.
21. Lechat, M.F. An epidemiologist's view of earth- Berkeley, CA, Earthquake Engineering Research
quakes. In: Solnes, J., ed. Engineering seismology Institute, 1983, pp. 277-279.
and earthquake engineering. Leiden, Noordhoff, 34. A study of earthquake losses in the San Francisco
1974, pp. 285-306 (NATO Advanced Study Bay Region. Prepared for the Office of Emergency
Institutes Series) Preparedness by the U.S. Department of Com-
22. The December 7, 1988, Armenia, USSR earth- merce, National Oceanic and Atmospheric
quake. San Francisco, CA, EQE Engineering, 1989, Administration, 1972.
pp. 1-48 (EQE Summary Report). 35. A study of earthquake losses in the Los Angeles,
23. Noji, E.K. Need for a sound research program on California, area. Prepared for the Federal Disaster
earthquake epidemiology. In: Hays, W.W., ed. Assistance Administration by the Department of
Proceedings of a Meeting of the U.S. Working Housing and Urban Development/U.S. Department
Group on Earthquake Related Casualties. Reston, of Commerce, National Oceanic and Atmospheric
VA, 13 July 1989. U.S. Geological Survey, 1990, Administration, 1973, pp. 1-331.
pp. 71-80 (Open-file Report 90-244). 36. Earthquake-associated deaths-California. Morbidity
24. Pollander, G.S. & Rund, D.A. Analysis of medical and mortality weekly report, 38: 767-770 (1990).
needs in disasters caused by earthquakes: the need 37. Clapperton, C.M. Volcanic and earthquake disas-
for a uniform injury reporting scheme. Disasters, 13: ters. Leicester, British Association for the
365-369 (1989). Advancement of Science, 1972, pp. 1-68.
25. Lechat, M.F. Updates: the epidemiology of health 38. Hogg, S.J. Reconstruction following seismic disas-
effects of disasters. Epidemiologic reviews, 12: ter in Venzone Friuli. Disasters, 4: 173-186 (1980).
192-197 (1990). 39. Tiedemann, H. Casualties as a function of building
26. Earthquake disaster - Luzon, Philippines. Morbidity quality and earthquake intensity. In: Jones, N.P. et
and mortality weekly report, 39: 573-577 (1990). al., ed. Proceedings of the International Workshop
27. Durkin, M.E. et al. The role of the physical setting on Earthquake Injury Epidemiology: Implications for
in earthquake injuries: the Mexico experience. In: Mitigations and Response, Baltimore, MD, 10-12
Bertero, V.V., ed. Reducing earthquake hazards: July 1989. Baltimore, MD, Johns Hopkins University,
lessons learned from the 1985 Mexico earthquake. 1989, pp. 420-434.
El Cerrito, CA, Earthquake Engineering Research 40. Hays, W.W. Proceedings of a Meeting of the U.S.
Institute, 1989, pp. 205-208. and ad hoc Working Group on Earthquake-related
28. Coburn, A.W. & Hughes, R.E. Fatalities, injury and Casualties, Reston, VA, 13 July 1989. Reston, VA,
rescue in earthquakes. In: Proceedings of the U.S. Geological Survey, 1990, pp. 1-147 (Open-file
Second Conference of the Development Studies Report 90-244).
Association, Manchester, England. Manchester 41. Jones, N.P. et al. Considerations in the epidemiolo-
University Press, 1987, pp. 235-243. gy of earthquake injuries. Earthquake spectra, 6:
29. Durkin, M.E. & Marakami, H.O. Casualties, survival 507-528 (1990).
and entrapment in heavy damaged buildings. In: 42. Meli, R. Modes of failure of buildings under seismic
Proceedings of the Ninth World Conference on actions. In: Jones, N.P. et al., ed. Proceedings of
Earthquake Engineering, Tokyo, Japan. Vol. 8. the International Workshop on Earthquake Injury
Tokyo, Japan Association for Earthquake Disaster Epidemiology: Implications for Mitigation and
Prevention, 1988, pp. 977-982. Response, Baltimore, MD, 10-12 July 1989.
30. Noji, E.K. et al. The 1988 earthquake in Soviet Baltimore, MD, Johns Hopkins University, 1989, pp.
Armenia: a case study. Annals of emergency medi- 366-377.
WHO Bulletin OMS. Vol 70 1992 257